A Peer-Based Approach on Analyzing Hacked Twitter Accounts

Abstract

Social media has become an important part of the lives of their hundreds of millions of users. Hackers make use of the large target audience by sending malicious content, often by hijacking existing accounts. This phenomenon has caused widespread research on how to detect hacked accounts, where different approaches exist. This work sets out to analyze the possibilities of including the reactions of hacked Twitter accounts’ peers into a detection system. Based on a dataset of six million tweets crawled from Twitter over the course of two years, we select a subset of tweets in which users react to alleged hacks of other accounts. We then gather and analyze the responses to those messages to reconstruct the conversations made. A quantitative analysis of these conversations shows that 30% of the users that are allegedly being hacked reply to the accusations, suggesting that these users acknowledge that their account was hacked.